A Multi-Coordination based Process for the design of mobile agent interactions

Mobile agent-based applications consist of autonomous, proactive and mobile software entities (or mobile agents) which interact to fulfil specific tasks. Mobile agent interactions should therefore be carefully designed to allow for both effectiveness and efficiency. Several techniques based on interaction design patterns and coordination models oriented to mobile agents have been to date proposed which can be used to design mobile agent interactions. However, an automated design and efficient coordination solutions is still lacking. This paper proposes the multi-coordination based process (MCP) for the design of mobile agent interactions which provides an effective coordination solution starting from a set of application-specific agent coordination requirements and properties. In particular, MCP is iterative and based on two subsequent phases of modeling and evaluation. The former uses interaction patterns and coordination models to semi-automatically provide alternative coordination solutions whereas the latter relies on simulation to evaluate and compare such solutions on the basis of ad-hoc defined performance indices. MCP is exemplified through a simple yet effective case study.